Major AI companies are buying millions of used books through intermediaries and destroying them after digitization to obtain high-quality training data while avoiding public scrutiny, according to a recent investigative report.
Tom’s Hardware reports that an investigation has revealed that AI companies are reportedly engaging in large-scale purchases of secondhand books to feed their language models, with the physical books subsequently being destroyed after scanning. The practice involves using middlemen to secretly acquire the books, potentially to avoid the controversy that has surrounded AI companies’ use of copyrighted material.
The demand for quality training data has pushed AI developers to seek out human-authored content that predates the proliferation of AI-generated material online. Books published before 2022 are particularly valuable because they are more likely to contain original content untainted by what has become known as “AI slop,” the low-quality AI-generated text now widespread across the internet.
ISBNdb, an online database containing over 111 million cataloged books, has reportedly shifted its business model to accommodate bulk book purchases for AI companies. According to the investigation, orders range from 1,000 copies to as many as one million books in single transactions.
Individual booksellers have noticed dramatic changes in their sales patterns. One anonymous professional bookseller told 404 Media that their weekly sales increased fivefold beginning in April of this year. Previously selling around 20 books in a good week, the seller now moves several hundred books weekly. Similar spikes have been reported by sellers on platforms including Alibris and Biblio.
Several characteristics of these purchases suggest AI company involvement. Buyers only request books with International Standard Book Numbers, the unique 13-digit codes used globally to identify publications. The purchases show no discernible pattern regarding subject matter, genre, or author. Perhaps most tellingly, purchasers appear indifferent to pricing, acquiring books regardless of cost, even when titles are overpriced.
The practice has legal precedent. Anthropic reportedly invested millions of dollars extracting information from countless printed books to build its Claude AI models before destroying the physical copies. The company sourced books from Better World Books. While a court decision determined that using books for AI training constitutes fair use under copyright law, Anthropic faced a $1.5 billion fine for maintaining a repository of seven million pirated books that infringed authors’ and publishers’ copyrights.
Google has faced similar legal challenges. A coalition of publishers recently filed a lawsuit accusing the tech giant of illegally using millions of copyrighted books to develop its Gemini AI models.
During the Anthropic lawsuit, Tom Harvey, who previously worked on Google Books before leading Anthropic’s “Project Panama” digitalization initiative, confirmed that the AI firm contracted several document scanning companies. Datamation Information Services was among them, offering both non-destructive and destructive book scanning services.
Non-destructive methods employ overhead scanners, flatbed scanners, or V-shaped imaging systems. Destructive scanning, however, involves physically dismantling books and feeding individual pages through high-speed industrial scanners. AI companies reportedly favor the destructive approach because it proves more efficient and cost-effective, resulting in the destruction of millions of books.
The practice raises significant ethical concerns. Critics question the morality of removing books from circulation, particularly since it remains unclear whether AI companies distinguish between rare or out-of-print titles and common books during the digitalization process.
Another major issue involves accessibility. Scanned books enter private databases used exclusively for AI training, making them unavailable to the general public. While the practice may produce more sophisticated AI systems, it potentially denies future generations access to valuable literary and informational resources.
AI giants continue their desperate search for training material to feed their insatiable AI models. This race to train AI continues to impact America in unpredictable ways. Breitbart News Social Media Director Wynton Hall has written his instant bestseller Code Red: The Left, the Right, China, and the Race to Control AI to serve as the definitive guide on how the MAGA movement can create positions on AI that benefit humanity without handing control of our nation to the leftists of Silicon Valley or allowing the Chinese to take over the world.
Read more at Tom’s Hardware here.
Lucas Nolan is a reporter for Breitbart News covering issues of AI, free speech, and online censorship.